Uber is turning its app into a travel stack and its fleet into a data-collection layer for autonomous vehicles, chief product officer Sachin Kansal told TechCrunch, framing both moves as bets that pay off before the company ever has to build its own self-driving system. Uber One membership has reached 51 million and accounts for roughly half of bookings, while 1.5 billion trips per year already happen outside a user's home city — the numbers behind a hotels partnership with Expedia and a new concierge feature called shop for me. A six-month-old business unit, AV Labs, will equip hundreds of cars with sensors to collect millions of miles of driving data.
The travel push landed earlier this year with hotels on Uber, boat rentals in Europe, and expanded shopping features. Kansal said the theme was deliberate: rides get people to the airport, Uber Eats has quietly replaced room service for many travelers, and hotels close the loop. Members booking a $1,000 hotel stay get 10% back as Uber credits — a $100 kicker that funnels demand back into rides and delivery.
“Travel really is, in my opinion, the third leg of the stool — we had rides, then we added eats, and now we are adding travel.”— Sachin Kansal, Uber Chief Product Officer
The Expedia integration is deep — Uber built the entire UI itself — while boat rentals hand users off to a partner's booking flow. Kansal described the handoff model as a way to test categories before committing engineering resources, not a permanent architecture. The membership math, he said, works out to two to three delivery orders per month to break even on the Uber One fee, and the cross-sell between rides and delivery is measurable within existing members.
Key facts
- 01Uber One now has 51 million members and drives roughly half of all bookings on the platform.
- 021.5 billion Uber trips per year happen outside a user's home city, the basis for the new travel push.
- 03AV Labs will equip hundreds of vehicles with sensors to collect millions of miles of driving data for autonomy partners.
- 04Uber wound down its Phoenix Waymo pilot of about a dozen cars while scaling to hundreds of cars in Austin and Atlanta.
- 05Uber handles 25 million lost items per year and has 10 million earners on the platform.
Uber Eats, long a drag on tech-sector profitability, is now standalone profitable and has been for several quarters, Kansal said. That matters because it removes the argument that ride-hailing is subsidizing delivery, and it clears the way for Uber to expand into travel and financial services without cannibalizing a still-fragile business line. The Uber Pro debit card for drivers and couriers is live, merchant financial products are being piloted, and consumer financial services remain an open question.
The autonomy story is where the strategy gets sharper. Uber recently wound down a Phoenix Waymo pilot of about a dozen cars while scaling to hundreds of cars in Austin and Atlanta with the same partner. Kansal said the two companies mutually decided Phoenix no longer made sense, an unusually candid acknowledgment that Waymo is simultaneously supplier and rival in several markets.
AV Labs is Uber's answer to that awkwardness. The unit will deploy sensor-equipped vehicles through fleet partners, generating edge-case driving data that autonomy developers need beyond the P95 and P99 cases their own fleets already cover. Uber is layering on operational know-how from 10 million earners and the 25 million lost items it handles per year — the mundane logistics that a purely software-driven L4 system still cannot address.
On the data-labeling side, Uber is already selling services to generative AI companies, using its earner base to label images and transcribe audio. Kansal was explicit that drivers do not record rider conversations; the audio work happens off-trip, and earners are paid separately for it. AV Labs data-sharing arrangements with autonomy partners are still being figured out, he said.
AI-driven features are showing up on both sides of the marketplace. An earner assistant tells drivers where demand is heading — suggesting a five-mile move out of a slow South Bay pocket, for example — while an Eats grocery cart assistant builds orders from a spoken list. Rides now supports voice requests that specify passenger count and luggage. A fully agentic trip-planning experience is a stated ambition without a shipping date.
The competitive frame Kansal painted is broad but not central to how he spends his time. Lyft in the US, Didi and 99 in Latin America, Bolt and Ola elsewhere, plus DoorDash and Delivery Hero in food, all show up on the map. Uber's positioning against Airbnb, which now offers airport transfers through a partner, is a newer wrinkle that Uber is answering with its Expedia hotel product rather than a direct short-term-rental play.
The clearest limit on the strategy is execution risk in businesses Uber has not run before. Hotel bookings, financial services, and AV data licensing all involve partners with more experience and, in some cases, competing incentives. Kansal's line that Uber does not want to be everything to everyone reads as a hedge against the super-app comparison to Grab — a business model that has struggled to prove out margins even in Asia.
For the AI market, Uber's AV Labs pivot is the more interesting signal. A company with 10 million earners and a global fleet is offering itself as a data infrastructure layer for autonomy developers rather than competing to build the stack — a bet that owning distribution and edge-case data beats owning the model. If that proves right, expect other marketplace companies with physical-world footprints to make similar plays, and expect the Waymo relationship to keep getting more complicated as Uber grows the leverage that comes from being indispensable to demand.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
